BLUEO

From BlueObelisk

On this pages any Blue Obelisk member can report their findings on the solubility contest send out by UCC members in July 2008. Anything that contributes to the analysis of the problem may be reported:

• prediction models
• improved 3D coordinates
• input validation (e.g. atom typing)
• aggregating Open Data on these molecules
• results in reproducing outcomes others reported
• etc (anything you think will contribute)

... with the only restriction that it fits ODOSOS, and thus uses open source tools: no Excel, but OOCalc and R are fine. no CORINA, by smi23d is fine, etc etc, Open Data and Open Standards. No limitation on programming language, just make sure to use open standards to report outcomes (for example CML).

The point is not to make the best model, but a model that cannot statistically be distinguished from the best model. If we do best, we have the advantage that we do not have to explain variance in predictability statistics :)

I suggest we write up everything in a paper; the best ranked contributors are invited, but even if we don't make that, and even if we're not even in statistical reach of the invited contributors, I'd suggest we'd write up our Open Science ODOSOS experiences and submit it anyway.

Your results

Please add your results below, and please add your name, the exact procedure to reproduce your results, and date. You're invited to report details in blogs etc, and are encouraged to link to such. At the end (only in some 7 weeks), we compose the recipe to reproduce the best results, and will require at least two indepedent Blue Obelisk community members to reproduce the results.

SVN is down at this right moment, but I will set up a SVN repository where we can easily share data, scripts etc.

Example Results, 2008-07-31, Egon Willighagen

Just the template format for results, so that we can easily track things.